kohya-ss / kohya-ss/sd-scripts
T5XXL tokenizer deadlocks when num workers>0
Open
- Dominant language
- Python
- Stars
- 7.2k
- Forks
- 1.2k
- Avg merge
- 11m
- Merged PRs (30d)
- 2
Description
Raised in #1586.
Contributor guide
No contributing guide indexed for this repository
Research direction
Start with the context in issue #1586 and reproduce the T5XXL tokenizer with num_workers>0. Trace the worker interaction to identify the deadlock; done means tokenization completes without hanging when multiple workers are enabled.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100